Last month I introduced the topic of Six Sigma and the role of Decision Management in it.^[1]  I also introduced the typical Six Sigma project known as a DMAIC project (Design, Measure, Analyze, Improve, and Control).  If you have a decision-intensive process, especially a reasonably high volume process, you can integrate Decision Management into Six Sigma and use these two complimentary approaches together.

In order to show the effectiveness of using Decision Management in a process you need to know the effectiveness of the decisions in the current process (Define/Measure), define what acceptable decisions are and implement a new decision or decisions (Analyze/Improve), and monitor the new decisions in order to show their effectiveness (Control).  Let's consider the steps in turn:

Table 1.  Outline of the Decision Management tasks in a DMAIC project

Define

During the define phase you must identify the decisions within the process.  You need to adopt the techniques of decision discovery to find the hidden decision, micro decisions, and, of course, obvious decisions within the operational processes that are the focus of the project.  For each decision you need to establish what combination of manual and automated decision-making is used currently.  While, in theory, the definition of a process should also define the decisions within it, I have found that an explicit focus on decisions is necessary otherwise the process will simply have poorly defined "diamonds" that are too generic to be useful.

Measure

In the measure phase you need to consider the possible ways to measure a decision.  You need to see if the decision lends itself to a simple measure of defect such as an accepted claim that turns out to be fraudulent.  Failing that, you need to consider the use of the five dimensions of what is known as Decision Yield (Precision, Consistency, Agility, Speed, and Cost) as a framework for defining defects.  Regardless of the approach taken you will need to identify target defect rates and ways to measure these defects for the decisions independent of any measures of the process. Just as processes should be linked to the business objectives and KPIs of your business so should decisions.

Analyze

In the analysis phase you need to analyze the code and procedures used for decisions in the process and establish root causes for poor decisions.  These causes might be bad or missing data, a lack of insight about what the data is telling you, poorly followed procedures, or out of date code.  Where you have data you should also consider using analytic modeling techniques to understand historical trends in decision making.  This analysis can show when and perhaps why a decision started to go bad, as well as identify potentially useful predictors that will improve the decision.  For complex decisions you should consider decision modeling and simulation to better understand decisions that impact each other over the lifetime of a customer.

Improve

One of the most effective ways to improve a process that is dependent on business decision is to apply a business rules management system to automate decisions.  This will give you an immediate improvement by replacing out-of-date code, poorly implemented or understood procedures, and other logic-related root causes with rules-driven decision-making.  It will also put in place the infrastructure you need to improve control, the next phase, and ensure that the next DMAIC project that focuses on this process will be able to rapidly and effectively improve the decisions within it.  Analytic techniques also come into play in improvement as you can use data mining techniques to find statistically-significant business rules and then deploy them using the BRMS.  These rules will be more effective than purely judgmental rules as they are systematically based on what has worked, and not worked, in the past.

Control

The final phase is that of control.  In this phase you should consider champion/challenger techniques, so you consider alternative rules and models to see which will impact defects most effectively.  You should also ensure that business users can control the business rules.  This is a key element of Six Sigma and a BRMS allows you to empower business users to manage the decisions within the process.  This is something that would fall under Voice of the Customer (VOC) in Six Sigma and will improve turn-around times, reduce maintenance work, and put business experts in the driving seat.  More advanced analytic techniques can be used to build decision-models to simulate the impact of changes in decisions on subsequent decisions and you can use these models to simulate potential changes and optimize decisions given real-world constraints.  Obviously, all this control builds on the ability to measure and analyze decisions and decision results.

A Six Sigma approach can and should be integrated with the tools, technologies, and approaches of Decision Management when you are applying Six Sigma to decision-centric processes.  While this is most true of processes that are about information products, the range of processes where some of the output can be described as information-based is wide so don't under call the number of processes where Decision Management can complement Six Sigma.

Don't forget that there is a lot more information on this in previous articles in this column as well as in my book Smart (Enough) Systems.  You can also use RSS or e-mail to subscribe to my blog (note new RSS and blog URLs).

References

[1] James Taylor, "Six Sigma and Decision Management — Introduction," Business Rules Journal, Vol. 10, No. 5 (May 2009), URL:  http://www.BRCommunity.com/a2009/b480.html  

# # #